edsl 0.1.39.dev3__py3-none-any.whl → 0.1.39.dev5__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (341) hide show
  1. edsl/Base.py +413 -332
  2. edsl/BaseDiff.py +260 -260
  3. edsl/TemplateLoader.py +24 -24
  4. edsl/__init__.py +57 -49
  5. edsl/__version__.py +1 -1
  6. edsl/agents/Agent.py +1071 -867
  7. edsl/agents/AgentList.py +551 -413
  8. edsl/agents/Invigilator.py +284 -233
  9. edsl/agents/InvigilatorBase.py +257 -270
  10. edsl/agents/PromptConstructor.py +272 -354
  11. edsl/agents/QuestionInstructionPromptBuilder.py +128 -0
  12. edsl/agents/QuestionTemplateReplacementsBuilder.py +137 -0
  13. edsl/agents/__init__.py +2 -3
  14. edsl/agents/descriptors.py +99 -99
  15. edsl/agents/prompt_helpers.py +129 -129
  16. edsl/agents/question_option_processor.py +172 -0
  17. edsl/auto/AutoStudy.py +130 -117
  18. edsl/auto/StageBase.py +243 -230
  19. edsl/auto/StageGenerateSurvey.py +178 -178
  20. edsl/auto/StageLabelQuestions.py +125 -125
  21. edsl/auto/StagePersona.py +61 -61
  22. edsl/auto/StagePersonaDimensionValueRanges.py +88 -88
  23. edsl/auto/StagePersonaDimensionValues.py +74 -74
  24. edsl/auto/StagePersonaDimensions.py +69 -69
  25. edsl/auto/StageQuestions.py +74 -73
  26. edsl/auto/SurveyCreatorPipeline.py +21 -21
  27. edsl/auto/utilities.py +218 -224
  28. edsl/base/Base.py +279 -279
  29. edsl/config.py +177 -157
  30. edsl/conversation/Conversation.py +290 -290
  31. edsl/conversation/car_buying.py +59 -58
  32. edsl/conversation/chips.py +95 -95
  33. edsl/conversation/mug_negotiation.py +81 -81
  34. edsl/conversation/next_speaker_utilities.py +93 -93
  35. edsl/coop/CoopFunctionsMixin.py +15 -0
  36. edsl/coop/ExpectedParrotKeyHandler.py +125 -0
  37. edsl/coop/PriceFetcher.py +54 -54
  38. edsl/coop/__init__.py +2 -2
  39. edsl/coop/coop.py +1106 -1028
  40. edsl/coop/utils.py +131 -131
  41. edsl/data/Cache.py +573 -555
  42. edsl/data/CacheEntry.py +230 -233
  43. edsl/data/CacheHandler.py +168 -149
  44. edsl/data/RemoteCacheSync.py +186 -78
  45. edsl/data/SQLiteDict.py +292 -292
  46. edsl/data/__init__.py +5 -4
  47. edsl/data/orm.py +10 -10
  48. edsl/data_transfer_models.py +74 -73
  49. edsl/enums.py +202 -175
  50. edsl/exceptions/BaseException.py +21 -21
  51. edsl/exceptions/__init__.py +54 -54
  52. edsl/exceptions/agents.py +54 -42
  53. edsl/exceptions/cache.py +5 -5
  54. edsl/exceptions/configuration.py +16 -16
  55. edsl/exceptions/coop.py +10 -10
  56. edsl/exceptions/data.py +14 -14
  57. edsl/exceptions/general.py +34 -34
  58. edsl/exceptions/inference_services.py +5 -0
  59. edsl/exceptions/jobs.py +33 -33
  60. edsl/exceptions/language_models.py +63 -63
  61. edsl/exceptions/prompts.py +15 -15
  62. edsl/exceptions/questions.py +109 -91
  63. edsl/exceptions/results.py +29 -29
  64. edsl/exceptions/scenarios.py +29 -22
  65. edsl/exceptions/surveys.py +37 -37
  66. edsl/inference_services/AnthropicService.py +106 -87
  67. edsl/inference_services/AvailableModelCacheHandler.py +184 -0
  68. edsl/inference_services/AvailableModelFetcher.py +215 -0
  69. edsl/inference_services/AwsBedrock.py +118 -120
  70. edsl/inference_services/AzureAI.py +215 -217
  71. edsl/inference_services/DeepInfraService.py +18 -18
  72. edsl/inference_services/GoogleService.py +143 -148
  73. edsl/inference_services/GroqService.py +20 -20
  74. edsl/inference_services/InferenceServiceABC.py +80 -147
  75. edsl/inference_services/InferenceServicesCollection.py +138 -97
  76. edsl/inference_services/MistralAIService.py +120 -123
  77. edsl/inference_services/OllamaService.py +18 -18
  78. edsl/inference_services/OpenAIService.py +236 -224
  79. edsl/inference_services/PerplexityService.py +160 -163
  80. edsl/inference_services/ServiceAvailability.py +135 -0
  81. edsl/inference_services/TestService.py +90 -89
  82. edsl/inference_services/TogetherAIService.py +172 -170
  83. edsl/inference_services/data_structures.py +134 -0
  84. edsl/inference_services/models_available_cache.py +118 -118
  85. edsl/inference_services/rate_limits_cache.py +25 -25
  86. edsl/inference_services/registry.py +41 -41
  87. edsl/inference_services/write_available.py +10 -10
  88. edsl/jobs/AnswerQuestionFunctionConstructor.py +223 -0
  89. edsl/jobs/Answers.py +43 -56
  90. edsl/jobs/FetchInvigilator.py +47 -0
  91. edsl/jobs/InterviewTaskManager.py +98 -0
  92. edsl/jobs/InterviewsConstructor.py +50 -0
  93. edsl/jobs/Jobs.py +823 -898
  94. edsl/jobs/JobsChecks.py +172 -147
  95. edsl/jobs/JobsComponentConstructor.py +189 -0
  96. edsl/jobs/JobsPrompts.py +270 -268
  97. edsl/jobs/JobsRemoteInferenceHandler.py +311 -239
  98. edsl/jobs/JobsRemoteInferenceLogger.py +239 -0
  99. edsl/jobs/RequestTokenEstimator.py +30 -0
  100. edsl/jobs/__init__.py +1 -1
  101. edsl/jobs/async_interview_runner.py +138 -0
  102. edsl/jobs/buckets/BucketCollection.py +104 -63
  103. edsl/jobs/buckets/ModelBuckets.py +65 -65
  104. edsl/jobs/buckets/TokenBucket.py +283 -251
  105. edsl/jobs/buckets/TokenBucketAPI.py +211 -0
  106. edsl/jobs/buckets/TokenBucketClient.py +191 -0
  107. edsl/jobs/check_survey_scenario_compatibility.py +85 -0
  108. edsl/jobs/data_structures.py +120 -0
  109. edsl/jobs/decorators.py +35 -0
  110. edsl/jobs/interviews/Interview.py +396 -661
  111. edsl/jobs/interviews/InterviewExceptionCollection.py +99 -99
  112. edsl/jobs/interviews/InterviewExceptionEntry.py +186 -186
  113. edsl/jobs/interviews/InterviewStatistic.py +63 -63
  114. edsl/jobs/interviews/InterviewStatisticsCollection.py +25 -25
  115. edsl/jobs/interviews/InterviewStatusDictionary.py +78 -78
  116. edsl/jobs/interviews/InterviewStatusLog.py +92 -92
  117. edsl/jobs/interviews/ReportErrors.py +66 -66
  118. edsl/jobs/interviews/interview_status_enum.py +9 -9
  119. edsl/jobs/jobs_status_enums.py +9 -0
  120. edsl/jobs/loggers/HTMLTableJobLogger.py +304 -0
  121. edsl/jobs/results_exceptions_handler.py +98 -0
  122. edsl/jobs/runners/JobsRunnerAsyncio.py +151 -466
  123. edsl/jobs/runners/JobsRunnerStatus.py +297 -330
  124. edsl/jobs/tasks/QuestionTaskCreator.py +244 -242
  125. edsl/jobs/tasks/TaskCreators.py +64 -64
  126. edsl/jobs/tasks/TaskHistory.py +470 -450
  127. edsl/jobs/tasks/TaskStatusLog.py +23 -23
  128. edsl/jobs/tasks/task_status_enum.py +161 -163
  129. edsl/jobs/tokens/InterviewTokenUsage.py +27 -27
  130. edsl/jobs/tokens/TokenUsage.py +34 -34
  131. edsl/language_models/ComputeCost.py +63 -0
  132. edsl/language_models/LanguageModel.py +626 -668
  133. edsl/language_models/ModelList.py +164 -155
  134. edsl/language_models/PriceManager.py +127 -0
  135. edsl/language_models/RawResponseHandler.py +106 -0
  136. edsl/language_models/RegisterLanguageModelsMeta.py +184 -184
  137. edsl/language_models/ServiceDataSources.py +0 -0
  138. edsl/language_models/__init__.py +2 -3
  139. edsl/language_models/fake_openai_call.py +15 -15
  140. edsl/language_models/fake_openai_service.py +61 -61
  141. edsl/language_models/key_management/KeyLookup.py +63 -0
  142. edsl/language_models/key_management/KeyLookupBuilder.py +273 -0
  143. edsl/language_models/key_management/KeyLookupCollection.py +38 -0
  144. edsl/language_models/key_management/__init__.py +0 -0
  145. edsl/language_models/key_management/models.py +131 -0
  146. edsl/language_models/model.py +256 -0
  147. edsl/language_models/repair.py +156 -156
  148. edsl/language_models/utilities.py +65 -64
  149. edsl/notebooks/Notebook.py +263 -258
  150. edsl/notebooks/NotebookToLaTeX.py +142 -0
  151. edsl/notebooks/__init__.py +1 -1
  152. edsl/prompts/Prompt.py +352 -362
  153. edsl/prompts/__init__.py +2 -2
  154. edsl/questions/ExceptionExplainer.py +77 -0
  155. edsl/questions/HTMLQuestion.py +103 -0
  156. edsl/questions/QuestionBase.py +518 -664
  157. edsl/questions/QuestionBasePromptsMixin.py +221 -217
  158. edsl/questions/QuestionBudget.py +227 -227
  159. edsl/questions/QuestionCheckBox.py +359 -359
  160. edsl/questions/QuestionExtract.py +180 -182
  161. edsl/questions/QuestionFreeText.py +113 -114
  162. edsl/questions/QuestionFunctional.py +166 -166
  163. edsl/questions/QuestionList.py +223 -231
  164. edsl/questions/QuestionMatrix.py +265 -0
  165. edsl/questions/QuestionMultipleChoice.py +330 -286
  166. edsl/questions/QuestionNumerical.py +151 -153
  167. edsl/questions/QuestionRank.py +314 -324
  168. edsl/questions/Quick.py +41 -41
  169. edsl/questions/SimpleAskMixin.py +74 -73
  170. edsl/questions/__init__.py +27 -26
  171. edsl/questions/{AnswerValidatorMixin.py → answer_validator_mixin.py} +334 -289
  172. edsl/questions/compose_questions.py +98 -98
  173. edsl/questions/data_structures.py +20 -0
  174. edsl/questions/decorators.py +21 -21
  175. edsl/questions/derived/QuestionLikertFive.py +76 -76
  176. edsl/questions/derived/QuestionLinearScale.py +90 -87
  177. edsl/questions/derived/QuestionTopK.py +93 -93
  178. edsl/questions/derived/QuestionYesNo.py +82 -82
  179. edsl/questions/descriptors.py +427 -413
  180. edsl/questions/loop_processor.py +149 -0
  181. edsl/questions/prompt_templates/question_budget.jinja +13 -13
  182. edsl/questions/prompt_templates/question_checkbox.jinja +32 -32
  183. edsl/questions/prompt_templates/question_extract.jinja +11 -11
  184. edsl/questions/prompt_templates/question_free_text.jinja +3 -3
  185. edsl/questions/prompt_templates/question_linear_scale.jinja +11 -11
  186. edsl/questions/prompt_templates/question_list.jinja +17 -17
  187. edsl/questions/prompt_templates/question_multiple_choice.jinja +33 -33
  188. edsl/questions/prompt_templates/question_numerical.jinja +36 -36
  189. edsl/questions/{QuestionBaseGenMixin.py → question_base_gen_mixin.py} +168 -161
  190. edsl/questions/question_registry.py +177 -177
  191. edsl/questions/{RegisterQuestionsMeta.py → register_questions_meta.py} +71 -71
  192. edsl/questions/{ResponseValidatorABC.py → response_validator_abc.py} +188 -174
  193. edsl/questions/response_validator_factory.py +34 -0
  194. edsl/questions/settings.py +12 -12
  195. edsl/questions/templates/budget/answering_instructions.jinja +7 -7
  196. edsl/questions/templates/budget/question_presentation.jinja +7 -7
  197. edsl/questions/templates/checkbox/answering_instructions.jinja +10 -10
  198. edsl/questions/templates/checkbox/question_presentation.jinja +22 -22
  199. edsl/questions/templates/extract/answering_instructions.jinja +7 -7
  200. edsl/questions/templates/likert_five/answering_instructions.jinja +10 -10
  201. edsl/questions/templates/likert_five/question_presentation.jinja +11 -11
  202. edsl/questions/templates/linear_scale/answering_instructions.jinja +5 -5
  203. edsl/questions/templates/linear_scale/question_presentation.jinja +5 -5
  204. edsl/questions/templates/list/answering_instructions.jinja +3 -3
  205. edsl/questions/templates/list/question_presentation.jinja +5 -5
  206. edsl/questions/templates/matrix/__init__.py +1 -0
  207. edsl/questions/templates/matrix/answering_instructions.jinja +5 -0
  208. edsl/questions/templates/matrix/question_presentation.jinja +20 -0
  209. edsl/questions/templates/multiple_choice/answering_instructions.jinja +9 -9
  210. edsl/questions/templates/multiple_choice/question_presentation.jinja +11 -11
  211. edsl/questions/templates/numerical/answering_instructions.jinja +6 -6
  212. edsl/questions/templates/numerical/question_presentation.jinja +6 -6
  213. edsl/questions/templates/rank/answering_instructions.jinja +11 -11
  214. edsl/questions/templates/rank/question_presentation.jinja +15 -15
  215. edsl/questions/templates/top_k/answering_instructions.jinja +8 -8
  216. edsl/questions/templates/top_k/question_presentation.jinja +22 -22
  217. edsl/questions/templates/yes_no/answering_instructions.jinja +6 -6
  218. edsl/questions/templates/yes_no/question_presentation.jinja +11 -11
  219. edsl/results/CSSParameterizer.py +108 -108
  220. edsl/results/Dataset.py +587 -424
  221. edsl/results/DatasetExportMixin.py +594 -731
  222. edsl/results/DatasetTree.py +295 -275
  223. edsl/results/MarkdownToDocx.py +122 -0
  224. edsl/results/MarkdownToPDF.py +111 -0
  225. edsl/results/Result.py +557 -465
  226. edsl/results/Results.py +1183 -1165
  227. edsl/results/ResultsExportMixin.py +45 -43
  228. edsl/results/ResultsGGMixin.py +121 -121
  229. edsl/results/TableDisplay.py +125 -198
  230. edsl/results/TextEditor.py +50 -0
  231. edsl/results/__init__.py +2 -2
  232. edsl/results/file_exports.py +252 -0
  233. edsl/results/{ResultsFetchMixin.py → results_fetch_mixin.py} +33 -33
  234. edsl/results/{Selector.py → results_selector.py} +145 -135
  235. edsl/results/{ResultsToolsMixin.py → results_tools_mixin.py} +98 -98
  236. edsl/results/smart_objects.py +96 -0
  237. edsl/results/table_data_class.py +12 -0
  238. edsl/results/table_display.css +77 -77
  239. edsl/results/table_renderers.py +118 -0
  240. edsl/results/tree_explore.py +115 -115
  241. edsl/scenarios/ConstructDownloadLink.py +109 -0
  242. edsl/scenarios/DocumentChunker.py +102 -0
  243. edsl/scenarios/DocxScenario.py +16 -0
  244. edsl/scenarios/FileStore.py +511 -632
  245. edsl/scenarios/PdfExtractor.py +40 -0
  246. edsl/scenarios/Scenario.py +498 -601
  247. edsl/scenarios/ScenarioHtmlMixin.py +65 -64
  248. edsl/scenarios/ScenarioList.py +1458 -1287
  249. edsl/scenarios/ScenarioListExportMixin.py +45 -52
  250. edsl/scenarios/ScenarioListPdfMixin.py +239 -261
  251. edsl/scenarios/__init__.py +3 -4
  252. edsl/scenarios/directory_scanner.py +96 -0
  253. edsl/scenarios/file_methods.py +85 -0
  254. edsl/scenarios/handlers/__init__.py +13 -0
  255. edsl/scenarios/handlers/csv.py +38 -0
  256. edsl/scenarios/handlers/docx.py +76 -0
  257. edsl/scenarios/handlers/html.py +37 -0
  258. edsl/scenarios/handlers/json.py +111 -0
  259. edsl/scenarios/handlers/latex.py +5 -0
  260. edsl/scenarios/handlers/md.py +51 -0
  261. edsl/scenarios/handlers/pdf.py +68 -0
  262. edsl/scenarios/handlers/png.py +39 -0
  263. edsl/scenarios/handlers/pptx.py +105 -0
  264. edsl/scenarios/handlers/py.py +294 -0
  265. edsl/scenarios/handlers/sql.py +313 -0
  266. edsl/scenarios/handlers/sqlite.py +149 -0
  267. edsl/scenarios/handlers/txt.py +33 -0
  268. edsl/scenarios/{ScenarioJoin.py → scenario_join.py} +131 -127
  269. edsl/scenarios/scenario_selector.py +156 -0
  270. edsl/shared.py +1 -1
  271. edsl/study/ObjectEntry.py +173 -173
  272. edsl/study/ProofOfWork.py +113 -113
  273. edsl/study/SnapShot.py +80 -80
  274. edsl/study/Study.py +521 -528
  275. edsl/study/__init__.py +4 -4
  276. edsl/surveys/ConstructDAG.py +92 -0
  277. edsl/surveys/DAG.py +148 -148
  278. edsl/surveys/EditSurvey.py +221 -0
  279. edsl/surveys/InstructionHandler.py +100 -0
  280. edsl/surveys/Memory.py +31 -31
  281. edsl/surveys/MemoryManagement.py +72 -0
  282. edsl/surveys/MemoryPlan.py +244 -244
  283. edsl/surveys/Rule.py +327 -326
  284. edsl/surveys/RuleCollection.py +385 -387
  285. edsl/surveys/RuleManager.py +172 -0
  286. edsl/surveys/Simulator.py +75 -0
  287. edsl/surveys/Survey.py +1280 -1801
  288. edsl/surveys/SurveyCSS.py +273 -261
  289. edsl/surveys/SurveyExportMixin.py +259 -259
  290. edsl/surveys/{SurveyFlowVisualizationMixin.py → SurveyFlowVisualization.py} +181 -179
  291. edsl/surveys/SurveyQualtricsImport.py +284 -284
  292. edsl/surveys/SurveyToApp.py +141 -0
  293. edsl/surveys/__init__.py +5 -3
  294. edsl/surveys/base.py +53 -53
  295. edsl/surveys/descriptors.py +60 -56
  296. edsl/surveys/instructions/ChangeInstruction.py +48 -49
  297. edsl/surveys/instructions/Instruction.py +56 -65
  298. edsl/surveys/instructions/InstructionCollection.py +82 -77
  299. edsl/templates/error_reporting/base.html +23 -23
  300. edsl/templates/error_reporting/exceptions_by_model.html +34 -34
  301. edsl/templates/error_reporting/exceptions_by_question_name.html +16 -16
  302. edsl/templates/error_reporting/exceptions_by_type.html +16 -16
  303. edsl/templates/error_reporting/interview_details.html +115 -115
  304. edsl/templates/error_reporting/interviews.html +19 -19
  305. edsl/templates/error_reporting/overview.html +4 -4
  306. edsl/templates/error_reporting/performance_plot.html +1 -1
  307. edsl/templates/error_reporting/report.css +73 -73
  308. edsl/templates/error_reporting/report.html +117 -117
  309. edsl/templates/error_reporting/report.js +25 -25
  310. edsl/tools/__init__.py +1 -1
  311. edsl/tools/clusters.py +192 -192
  312. edsl/tools/embeddings.py +27 -27
  313. edsl/tools/embeddings_plotting.py +118 -118
  314. edsl/tools/plotting.py +112 -112
  315. edsl/tools/summarize.py +18 -18
  316. edsl/utilities/PrettyList.py +56 -0
  317. edsl/utilities/SystemInfo.py +28 -28
  318. edsl/utilities/__init__.py +22 -22
  319. edsl/utilities/ast_utilities.py +25 -25
  320. edsl/utilities/data/Registry.py +6 -6
  321. edsl/utilities/data/__init__.py +1 -1
  322. edsl/utilities/data/scooter_results.json +1 -1
  323. edsl/utilities/decorators.py +77 -77
  324. edsl/utilities/gcp_bucket/cloud_storage.py +96 -96
  325. edsl/utilities/interface.py +627 -627
  326. edsl/utilities/is_notebook.py +18 -0
  327. edsl/utilities/is_valid_variable_name.py +11 -0
  328. edsl/utilities/naming_utilities.py +263 -263
  329. edsl/utilities/remove_edsl_version.py +24 -0
  330. edsl/utilities/repair_functions.py +28 -28
  331. edsl/utilities/restricted_python.py +70 -70
  332. edsl/utilities/utilities.py +436 -424
  333. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/LICENSE +21 -21
  334. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/METADATA +13 -11
  335. edsl-0.1.39.dev5.dist-info/RECORD +358 -0
  336. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/WHEEL +1 -1
  337. edsl/language_models/KeyLookup.py +0 -30
  338. edsl/language_models/registry.py +0 -190
  339. edsl/language_models/unused/ReplicateBase.py +0 -83
  340. edsl/results/ResultsDBMixin.py +0 -238
  341. edsl-0.1.39.dev3.dist-info/RECORD +0 -277
edsl/results/tree_explore.py CHANGED
@@ -1,115 +1,115 @@
1
- from collections import defaultdict
2
- from typing import List, Dict, Any
3
- import json
4
-
5
-
6
- class FoldableHTMLTableGenerator:
7
- def __init__(self, data: List[Dict[str, Any]]):
8
- self.data = data
9
-
10
- def tree(self, fold_attributes: List[str], drop: List[str] = None) -> Dict:
11
- def nested_dict():
12
- return defaultdict(nested_dict)
13
-
14
- result = nested_dict()
15
- drop = drop or [] # Use an empty list if drop is None
16
-
17
- for item in self.data:
18
- current = result
19
- for attr in fold_attributes:
20
- current = current[item[attr]]
21
-
22
- row = {
23
- k: v
24
- for k, v in item.items()
25
- if k not in fold_attributes and k not in drop
26
- }
27
- if "_rows" not in current:
28
- current["_rows"] = []
29
- current["_rows"].append(row)
30
-
31
- return result
32
-
33
- def generate_html(self, tree, fold_attributes: List[str]) -> str:
34
- html_content = """
35
- <!DOCTYPE html>
36
- <html lang="en">
37
- <head>
38
- <meta charset="UTF-8">
39
- <meta name="viewport" content="width=device-width, initial-scale=1.0">
40
- <title>Foldable Nested Table</title>
41
- <style>
42
- .folding-section { margin-left: 20px; }
43
- .fold-button { cursor: pointer; margin: 5px 0; }
44
- table { border-collapse: collapse; margin-top: 10px; }
45
- th, td { border: 1px solid black; padding: 5px; }
46
- .attribute-label { font-weight: bold; }
47
- </style>
48
- </head>
49
- <body>
50
- <div id="root"></div>
51
- <script>
52
- function toggleFold(id) {
53
- const element = document.getElementById(id);
54
- element.style.display = element.style.display === 'none' ? 'block' : 'none';
55
- }
56
-
57
- function createFoldableSection(data, path = [], attributes = %s) {
58
- const container = document.createElement('div');
59
- container.className = 'folding-section';
60
-
61
- for (const [key, value] of Object.entries(data)) {
62
- if (key === '_rows') {
63
- const table = document.createElement('table');
64
- const headerRow = table.insertRow();
65
- const headers = Object.keys(value[0]);
66
- headers.forEach(header => {
67
- const th = document.createElement('th');
68
- th.textContent = header;
69
- headerRow.appendChild(th);
70
- });
71
- value.forEach(row => {
72
- const tableRow = table.insertRow();
73
- headers.forEach(header => {
74
- const cell = tableRow.insertCell();
75
- cell.textContent = row[header];
76
- });
77
- });
78
- container.appendChild(table);
79
- } else {
80
- const button = document.createElement('button');
81
- const attributeType = attributes[path.length];
82
- button.innerHTML = `<span class="attribute-label">${attributeType}:</span> ${key}`;
83
- button.className = 'fold-button';
84
- const sectionId = `section-${path.join('-')}-${key}`;
85
- button.onclick = () => toggleFold(sectionId);
86
- container.appendChild(button);
87
-
88
- const section = document.createElement('div');
89
- section.id = sectionId;
90
- section.style.display = 'none';
91
- section.appendChild(createFoldableSection(value, [...path, key], attributes));
92
- container.appendChild(section);
93
- }
94
- }
95
-
96
- return container;
97
- }
98
-
99
- const treeData = %s;
100
- document.getElementById('root').appendChild(createFoldableSection(treeData));
101
- </script>
102
- </body>
103
- </html>
104
- """
105
-
106
- return html_content % (json.dumps(fold_attributes), json.dumps(tree))
107
-
108
- def save_html(self, fold_attributes: List[str], filename: str = "output.html"):
109
- tree = self.tree(fold_attributes)
110
- html_content = self.generate_html(tree, fold_attributes)
111
-
112
- with open(filename, "w", encoding="utf-8") as f:
113
- f.write(html_content)
114
-
115
- print(f"HTML file has been generated: {filename}")
1
+ from collections import defaultdict
2
+ from typing import List, Dict, Any
3
+ import json
4
+
5
+
6
+ class FoldableHTMLTableGenerator:
7
+ def __init__(self, data: List[Dict[str, Any]]):
8
+ self.data = data
9
+
10
+ def tree(self, fold_attributes: List[str], drop: List[str] = None) -> Dict:
11
+ def nested_dict():
12
+ return defaultdict(nested_dict)
13
+
14
+ result = nested_dict()
15
+ drop = drop or [] # Use an empty list if drop is None
16
+
17
+ for item in self.data:
18
+ current = result
19
+ for attr in fold_attributes:
20
+ current = current[item[attr]]
21
+
22
+ row = {
23
+ k: v
24
+ for k, v in item.items()
25
+ if k not in fold_attributes and k not in drop
26
+ }
27
+ if "_rows" not in current:
28
+ current["_rows"] = []
29
+ current["_rows"].append(row)
30
+
31
+ return result
32
+
33
+ def generate_html(self, tree, fold_attributes: List[str]) -> str:
34
+ html_content = """
35
+ <!DOCTYPE html>
36
+ <html lang="en">
37
+ <head>
38
+ <meta charset="UTF-8">
39
+ <meta name="viewport" content="width=device-width, initial-scale=1.0">
40
+ <title>Foldable Nested Table</title>
41
+ <style>
42
+ .folding-section { margin-left: 20px; }
43
+ .fold-button { cursor: pointer; margin: 5px 0; }
44
+ table { border-collapse: collapse; margin-top: 10px; }
45
+ th, td { border: 1px solid black; padding: 5px; }
46
+ .attribute-label { font-weight: bold; }
47
+ </style>
48
+ </head>
49
+ <body>
50
+ <div id="root"></div>
51
+ <script>
52
+ function toggleFold(id) {
53
+ const element = document.getElementById(id);
54
+ element.style.display = element.style.display === 'none' ? 'block' : 'none';
55
+ }
56
+
57
+ function createFoldableSection(data, path = [], attributes = %s) {
58
+ const container = document.createElement('div');
59
+ container.className = 'folding-section';
60
+
61
+ for (const [key, value] of Object.entries(data)) {
62
+ if (key === '_rows') {
63
+ const table = document.createElement('table');
64
+ const headerRow = table.insertRow();
65
+ const headers = Object.keys(value[0]);
66
+ headers.forEach(header => {
67
+ const th = document.createElement('th');
68
+ th.textContent = header;
69
+ headerRow.appendChild(th);
70
+ });
71
+ value.forEach(row => {
72
+ const tableRow = table.insertRow();
73
+ headers.forEach(header => {
74
+ const cell = tableRow.insertCell();
75
+ cell.textContent = row[header];
76
+ });
77
+ });
78
+ container.appendChild(table);
79
+ } else {
80
+ const button = document.createElement('button');
81
+ const attributeType = attributes[path.length];
82
+ button.innerHTML = `<span class="attribute-label">${attributeType}:</span> ${key}`;
83
+ button.className = 'fold-button';
84
+ const sectionId = `section-${path.join('-')}-${key}`;
85
+ button.onclick = () => toggleFold(sectionId);
86
+ container.appendChild(button);
87
+
88
+ const section = document.createElement('div');
89
+ section.id = sectionId;
90
+ section.style.display = 'none';
91
+ section.appendChild(createFoldableSection(value, [...path, key], attributes));
92
+ container.appendChild(section);
93
+ }
94
+ }
95
+
96
+ return container;
97
+ }
98
+
99
+ const treeData = %s;
100
+ document.getElementById('root').appendChild(createFoldableSection(treeData));
101
+ </script>
102
+ </body>
103
+ </html>
104
+ """
105
+
106
+ return html_content % (json.dumps(fold_attributes), json.dumps(tree))
107
+
108
+ def save_html(self, fold_attributes: List[str], filename: str = "output.html"):
109
+ tree = self.tree(fold_attributes)
110
+ html_content = self.generate_html(tree, fold_attributes)
111
+
112
+ with open(filename, "w", encoding="utf-8") as f:
113
+ f.write(html_content)
114
+
115
+ print(f"HTML file has been generated: {filename}")
edsl/scenarios/ConstructDownloadLink.py ADDED
@@ -0,0 +1,109 @@
1
+ import os
2
+ import mimetypes
3
+
4
+
5
+ class ConstructDownloadLink:
6
+ """
7
+ A class to create HTML download links for FileStore objects.
8
+ The links can be displayed in Jupyter notebooks or other web interfaces.
9
+ """
10
+
11
+ def __init__(self, filestore):
12
+ """
13
+ Initialize with a FileStore object.
14
+
15
+ Args:
16
+ filestore: A FileStore object containing the file to be made downloadable
17
+ """
18
+ self.filestore = filestore
19
+
20
+ def create_link(self, custom_filename=None, style=None):
21
+ from IPython.display import HTML
22
+
23
+ html = self.html_create_link(custom_filename, style)
24
+ return HTML(html)
25
+
26
+ def html_create_link(self, custom_filename=None, style=None):
27
+ """
28
+ Create an HTML download link for the file.
29
+
30
+ Args:
31
+ custom_filename (str, optional): Custom name for the downloaded file.
32
+ If None, uses original filename.
33
+ style (dict, optional): Custom CSS styles for the download button.
34
+ If None, uses default styling.
35
+
36
+ Returns:
37
+ IPython.display.HTML: HTML object containing the download link
38
+ """
39
+
40
+ # Get filename from path or use custom filename
41
+ original_filename = os.path.basename(self.filestore.path)
42
+ filename = custom_filename or original_filename
43
+
44
+ # Use the base64 string already stored in FileStore
45
+ b64_data = self.filestore.base64_string
46
+
47
+ # Use mime type from FileStore or guess it
48
+ mime_type = self.filestore.mime_type
49
+
50
+ # Default style if none provided
51
+ default_style = {
52
+ "background-color": "#4CAF50",
53
+ "color": "white",
54
+ "padding": "10px 20px",
55
+ "text-decoration": "none",
56
+ "border-radius": "4px",
57
+ "display": "inline-block",
58
+ "margin": "10px 0",
59
+ "font-family": "sans-serif",
60
+ "cursor": "pointer",
61
+ }
62
+
63
+ button_style = style or default_style
64
+ style_str = "; ".join(f"{k}: {v}" for k, v in button_style.items())
65
+
66
+ html = f"""
67
+ <a download="{filename}"
68
+ href="data:{mime_type};base64,{b64_data}"
69
+ style="{style_str}">
70
+ Download {filename}
71
+ </a>
72
+ """
73
+ return html
74
+
75
+ def create_multiple_links(self, files, custom_filenames=None, style=None):
76
+ """
77
+ Create multiple download links at once.
78
+ Useful when you want to provide different versions of the same file
79
+ or related files together.
80
+
81
+ Args:
82
+ files (list): List of FileStore objects
83
+ custom_filenames (list, optional): List of custom filenames for downloads
84
+ style (dict, optional): Custom CSS styles for the download buttons
85
+
86
+ Returns:
87
+ IPython.display.HTML: HTML object containing all download links
88
+ """
89
+ if custom_filenames is None:
90
+ custom_filenames = [None] * len(files)
91
+
92
+ html_parts = []
93
+ for file_obj, custom_name in zip(files, custom_filenames):
94
+ link_creator = ConstructDownloadLink(file_obj)
95
+ html_parts.append(
96
+ link_creator.create_link(
97
+ custom_filename=custom_name, style=style
98
+ )._repr_html_()
99
+ )
100
+
101
+ return HTML(
102
+ '<div style="display: flex; gap: 10px;">' + "".join(html_parts) + "</div>"
103
+ )
104
+
105
+
106
+ if __name__ == "__main__":
107
+ import doctest
108
+
109
+ doctest.testmod()
edsl/scenarios/DocumentChunker.py ADDED
@@ -0,0 +1,102 @@
1
+ from __future__ import annotations
2
+ from typing import Optional, Generator, TYPE_CHECKING
3
+ import copy
4
+
5
+ if TYPE_CHECKING:
6
+ from edsl.scenarios.Scenario import Scenario
7
+ from edsl.scenarios.ScenarioList import ScenarioList
8
+
9
+
10
+ class DocumentChunker:
11
+ def __init__(self, scenario: "Scenario"):
12
+ self.scenario = scenario
13
+
14
+ @staticmethod
15
+ def _line_chunks(text, num_lines: int) -> Generator[str, None, None]:
16
+ """Split a text into chunks of a given size.
17
+
18
+ :param text: The text to split.
19
+ :param num_lines: The number of lines in each chunk.
20
+
21
+ Example:
22
+
23
+ >>> list(DocumentChunker._line_chunks("This is a test.\\nThis is a test. This is a test.", 1))
24
+ ['This is a test.', 'This is a test. This is a test.']
25
+ """
26
+ lines = text.split("\n")
27
+ for i in range(0, len(lines), num_lines):
28
+ chunk = "\n".join(lines[i : i + num_lines])
29
+ yield chunk
30
+
31
+ @staticmethod
32
+ def _word_chunks(text, num_words: int) -> Generator[str, None, None]:
33
+ """Split a text into chunks of a given size.
34
+
35
+ :param text: The text to split.
36
+ :param num_words: The number of words in each chunk.
37
+
38
+ Example:
39
+
40
+ >>> list(DocumentChunker._word_chunks("This is a test.", 2))
41
+ ['This is', 'a test.']
42
+ """
43
+ words = text.split()
44
+ for i in range(0, len(words), num_words):
45
+ chunk = " ".join(words[i : i + num_words])
46
+ yield chunk
47
+
48
+ def chunk(
49
+ self,
50
+ field,
51
+ num_words: Optional[int] = None,
52
+ num_lines: Optional[int] = None,
53
+ include_original=False,
54
+ hash_original=False,
55
+ ) -> ScenarioList:
56
+ """Split a field into chunks of a given size.
57
+
58
+ :param field: The field to split.
59
+ :param num_words: The number of words in each chunk.
60
+ :param num_lines: The number of lines in each chunk.
61
+ :param include_original: Whether to include the original field in the new scenarios.
62
+ :param hash_original: Whether to hash the original field in the new scenarios.
63
+
64
+ If you specify `include_original=True`, the original field will be included in the new scenarios with an "_original" suffix.
65
+ """
66
+ from edsl.scenarios.ScenarioList import ScenarioList
67
+ import hashlib
68
+
69
+ if num_words is not None:
70
+ chunks = list(self._word_chunks(self.scenario[field], num_words))
71
+
72
+ if num_lines is not None:
73
+ chunks = list(self._line_chunks(self.scenario[field], num_lines))
74
+
75
+ if num_words is None and num_lines is None:
76
+ raise ValueError("You must specify either num_words or num_lines.")
77
+
78
+ if num_words is not None and num_lines is not None:
79
+ raise ValueError(
80
+ "You must specify either num_words or num_lines, but not both."
81
+ )
82
+
83
+ scenarios = []
84
+ for i, chunk in enumerate(chunks):
85
+ new_scenario = copy.deepcopy(self.scenario)
86
+ new_scenario[field] = chunk
87
+ new_scenario[field + "_chunk"] = i
88
+ if include_original:
89
+ if hash_original:
90
+ new_scenario[field + "_original"] = hashlib.md5(
91
+ self.scenario[field].encode()
92
+ ).hexdigest()
93
+ else:
94
+ new_scenario[field + "_original"] = self.scenario[field]
95
+ scenarios.append(new_scenario)
96
+ return ScenarioList(scenarios)
97
+
98
+
99
+ if __name__ == "__main__":
100
+ import doctest
101
+
102
+ doctest.testmod()
edsl/scenarios/DocxScenario.py ADDED
@@ -0,0 +1,16 @@
1
+ class DocxScenario:
2
+ def __init__(self, docx_path: str):
3
+ from docx import Document
4
+
5
+ self.doc = Document(docx_path)
6
+ self.docx_path = docx_path
7
+
8
+ def get_scenario_dict(self) -> dict:
9
+ # Extract all text
10
+ full_text = []
11
+ for para in self.doc.paragraphs:
12
+ full_text.append(para.text)
13
+
14
+ # Join the text from all paragraphs
15
+ text = "\n".join(full_text)
16
+ return {"file_path": self.docx_path, "text": text}